SI
SI
discoversearch

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor.  We ask that you disable ad blocking while on Silicon Investor in the best interests of our community.  If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.
Politics : Should God be replaced?

 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext  
To: cosmicforce who wrote (1038)9/23/2000 11:38:12 AM
From: TigerPaw   of 28931
 
discrete nature of the double precision math I used
The net treats the discrete nature like a defect and can learn around defects. You may have had the back propagation set too tight in an effort to train the net quicker. It has trouble sliding around the discontinuities with too strong feedback. A lot of what I did involved automatic training sessions so that examples could be retrained slowly but repeated millions of times before the training was complete. There is also a step I term simulated annealing which helps lift the net out of local minimums.
TP
Report TOU ViolationShare This Post
 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext